From nobody Wed Sep 29 13:17:01 2021 X-Original-To: fs@mlmmj.nyi.freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2610:1c1:1:606c::19:1]) by mlmmj.nyi.freebsd.org (Postfix) with ESMTP id 654B217CD325 for ; Wed, 29 Sep 2021 13:17:04 +0000 (UTC) (envelope-from avg@freebsd.org) Received: from smtp.freebsd.org (smtp.freebsd.org [IPv6:2610:1c1:1:606c::24b:4]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (4096 bits) server-digest SHA256 client-signature RSA-PSS (4096 bits) client-digest SHA256) (Client CN "smtp.freebsd.org", Issuer "R3" (verified OK)) by mx1.freebsd.org (Postfix) with ESMTPS id 4HKH342RYcz3jjP; Wed, 29 Sep 2021 13:17:04 +0000 (UTC) (envelope-from avg@freebsd.org) Received: from [192.168.0.88] (unknown [195.64.148.76]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (Client did not present a certificate) (Authenticated sender: avg/mail) by smtp.freebsd.org (Postfix) with ESMTPSA id D02BD29FEA; Wed, 29 Sep 2021 13:17:03 +0000 (UTC) (envelope-from avg@freebsd.org) From: Andriy Gapon Subject: Re: vfs.zfs.vol.recursive completely broken, time to remove? To: Alan Somers Cc: freebsd-fs References: <10583cba-c2f5-ffbe-5a1b-8dfcd8114b4d@FreeBSD.org> Message-ID: Date: Wed, 29 Sep 2021 16:17:01 +0300 User-Agent: Mozilla/5.0 (X11; FreeBSD amd64; rv:78.0) Gecko/20100101 Firefox/78.0 Thunderbird/78.14.0 List-Id: Filesystems List-Archive: https://lists.freebsd.org/archives/freebsd-fs List-Help: List-Post: List-Subscribe: List-Unsubscribe: Sender: owner-freebsd-fs@freebsd.org MIME-Version: 1.0 In-Reply-To: Content-Type: text/plain; charset=utf-8; format=flowed Content-Language: en-US Content-Transfer-Encoding: 7bit X-Spam: Yes X-ThisMailContainsUnwantedMimeParts: N On 29/09/2021 16:05, Alan Somers wrote: > On Wed, Sep 29, 2021 at 12:10 AM Andriy Gapon wrote: >> >> >> It seems that while previously there could sometimes be a deadlock with >> vfs.zfs.vol.recursive=1, now it is guaranteed. >> >> There was an OpenZFS change to use a pool of threads for vdev opening (to >> parallelize / speed up). When such a thread opens a zvol it needs to acquire >> spa_namespace_lock. But the lock is already held by a thread requesting the >> operation such as a thread doing zpool import (spa_load). >> So, it's an obvious deadlock: the main thread is waiting for the workers while >> holding the namespace lock, at least one of the workers is blocked on the same lock. > > The last time I tried that there was pushback from a lot of different > people. I think we need to fix it instead. Here's the beginning of > that flame war. > https://docs.freebsd.org/cgi/getmsg.cgi?fetch=1031189+0+archive/2016/svn-src-all/20160124.svn-src-all > Back then it probably had a chance of working. At least, I have been using it for a long time without much problems. Now (for the past half a year or so) it's completely unusable. -- Andriy Gapon